The fitness landscape of a community of Darwin's finches.

Divergent natural selection should lead to adaptive radiation - that is, the rapid evolution of phenotypic and ecological diversity originating from a single clade. The drivers of adaptive radiations have often been conceptualized through the concept of "adaptive landscapes"; yet formal empirical estimates of adaptive landscapes for natural adaptive radiations have proven elusive. Here we use a 17-year dataset of Darwin's ground finches (Geospiza spp.) at an intensively-studied site on Santa Cruz (Galápagos) to estimate individual apparent lifespan in relation to beak traits. We use these estimates to model a multi-species fitness landscape, which we also convert to a formal adaptive landscape. We then assess correspondence between estimated fitness peaks and observed phenotypes for each of five phenotypic modes (G. fuliginosa, G. fortis [small and large morphotypes], G. magnirostris, and G. scandens). The fitness and adaptive landscapes show five and four peaks, respectively; and, as expected, the adaptive landscape was smoother than the fitness landscape. Each of the five phenotypic modes appeared reasonably close to the corresponding fitness peak; yet interesting deviations were also documented and examined. By estimating adaptive landscapes in an ongoing adaptive radiation, our study demonstrates their utility as a quantitative tool for exploring and predicting adaptive radiation.